SMILE

SMILE 1.4 is now available. Notable new features in this release include:

  • full support for Python 3.x
  • new functions, random generators and custom functions for use in equation-based nodes
  • noisyMAX decomposition algorithm
  • specialized entropy-based diagnosis algorithm for mutually exclusive faults
  • performance and reliability improvements

Libraries are available for download now at https://download.bayesfusion.com.

Fun fact: during the full build process we currently compile 60 SMILE variants for use with Visual C++. 60 = 5 x 3 x 2 x 2.

  • 5 Visual Studio versions (2010, ’12, ’13, ’15, ’17)
  • 3 C++ runtime types (static release, DLL release, DLL debug)
  • 2 CPU architectures (x86 and x64)
  • 2 product editions (Academic and Business)

This of course is in addition to many versions for Linux, Mac, iOS and accompanying wrappers for multiple programming languages.

BayesFusion now offers free, 30-day, hosted evaluation of BayesBox. We maintain the server, the customer only needs to upload the networks using browser-based admin interface. Access to uploaded models can be protected by password on request.

Contact us for details.

BayesFusion releases GeNIe 2.4 today. Users can now define their own functions for use in continuous and hybrid Bayesian models. NoisyMAX and noisyAdder are now fully supported in hybrid models. It’s also possible to download models directly from a BayesBox instance.

We have released BayesMobile, an application for mobile devices which can be used to perform inference and diagnosis with Bayesian models.

We have released BayesBox, a software for using Bayesian networks in a web browser. To see it in action, visit our interactive model repository at https://repo.bayesfusion.com or a demo site at https://demo.bayesfusion.com.

BayesFusion releases GeNIe 2.3 with numerous enhancements to the user interface, especially in its learning functionality. Among other improvements, GeNIe 2.3 introduces direct treatment of Noisy-MAX gates in inference algorithms, which greatly improves performance for large models containing Noisy-MAX gates.

SMILE

SMILE is now available for Visual Studio 2017.

The Program Committee of the 9th Probabilistic Graphical Models (PGM-2018) conference announced the winners of the BayesFusion Best Student Paper Award during the PGM-2018 conference banquet in Prague, Czech Republic, on September 13.  The winners are:

BayesFusion Best Student Paper Award given jointly to:

Irene Córdoba, Department of Artificial Intelligence Universidad Politécnica de Madrid, Spain, for the paper entitled A Partial Orthogonalization Method for Simulating Covariance and Concentration Graph Matrices, co-authored with Gherardo Varando, Concha Bielza and Pedro Larrañaga

Kari Rantanen, HIIT, Department of Computer Science, University of Helsinki, Finland, for the paper entitled Learning Optimal Causal Graphs with Exact Search, co-authored with Antti Hyttinen and Matti Järvisalo

Honorable Mentions:
Gherardo Varando, Department of Artificial Intelligence Universidad Politécnica de Madrid, Spain, and Department of Mathematical Sciences, University of Copenhagen, Denmark, for being a student co-author of the paper that won the BayesFusion Best Student Paper Award, entitled A Partial Orthogonalization Method for Simulating Covariance and Concentration Graph Matrices, with Irene Córdoba, Concha Bielza and Pedro Larrañaga

Janne Leppä-aho, University of Helsinki, Department of Computer Science / HIIT, Finland, for the paper entitled Learning Non-parametric Markov Networks with Mutual Information, co-authored with Santeri Räisänen, Xiao Yang and Teemu Roos

SMILE

We have added iOS to the list of platforms for which SMILE is compiled. The library is available for download.

SMILE licenses generated by our licensing server include the license for use with R/rJava (in addition to C++, Python, Java, C# and VB.NET).